(d) A cell-centered AMR grid can still be transformed using dual cells however, stitching across the boundary remains challenging. At the level boundaries of vertex-centered AMR data (c), it is sufficient to introduce a layer of ghost cells. (b) A cell-centered single-level grid can be converted to a vertex-centered grid by introducing dual cells. (a) Trilinear interpolation is trivial on a vertex-centered single-level grid. Finally, we integrate our reconstruction method and hybrid implicit isosurface approach into the OSPRay ray-tracing framework as a module, allowing us to trivially support multiple transparent isosurfaces, combined isosurface and volume rendering and advanced shading effects. To render these isosurfaces, we combine ideas from isosurface extraction and implicit isosurface ray tracing and present an efficient hybrid implicit isosurface ray-tracing approach, which allows for semi-interactive changes to the isovalue. We build our approach on a novel reconstruction method for BS-AMR data, called the octant method, that allows us to construct crack-free implicit isosurfaces, even across level boundaries. In this paper, we propose an efficient solution for isosurface visualization of large-scale BS-AMR data. Thus, an efficient approach for direct isosurface visualization of AMR data on CPUs remains desirable, due to both the prevalence of CPUs on current and upcoming HPC systems and the large amount of memory available. GPU-based approaches for visualizing such data typically remain in special-purpose tools, and are limited by the size of the GPU memory, requiring data-parallel rendering to support the large datasets produced by current simulations. Prior work has proposed to introduce unstructured mesh elements to stitch across level boundaries, at the cost of requiring the rendering method to handle unstructured elements. Down-sampling the data clearly comes with an undesirable loss of resolution in regions of interest in the data, whereas up-sampling the data may require an exorbitant amount of memory.Ī key challenge in directly rendering AMR data is reconstructing the data at level boundaries. General visualization frameworks such as VTK, ParaView and VisIt provide limited support for direct visualization of AMR datasets, requiring the user to either down- or up-sample the data to a fixed resolution grid before rendering. Existing visualization solutions for large-scale AMR data remain either special purpose or have severe limitations. Īlthough BS-AMR techniques have found wide adoption in current large-scale HPC simulations, visualization techniques for such data have struggled to keep up. A detailed overview of these frameworks, and other BS-AMR-based simulations, can be found in Dubey et al.’s survey. BS-AMR forms the basis for a number of scientific simulation frameworks, including BoxLib, LAVA, Chombo, GR-Chombo, Enzo, AMReX and Uintah. Although other forms of AMR data exist (e.g., mesh distortion and tree-based), block-structured AMR (BS-AMR) is the most widely used in practice, as it can be easily coupled with octree or recursive-grid AMR. By providing an adaptive, hierarchical resolution representation of the computational domain, AMR techniques allow the simulation to focus both computational effort and storage on regions of interest, enabling larger, more complex problems to be solved. We evaluate the rendering performance, memory consumption and quality of our method on two gigascale block-structured AMR datasets.Īdaptive mesh refinement (AMR) techniques are used to solve a range of complex problems in numerical analysis. Finally, we integrate our octant method and hybrid isosurface geometry into OSPRay as a module, providing the ability to create high-quality interactive visualizations combining volume and isosurface representations of BS-AMR data. Furthermore, we present a generally applicable hybrid implicit isosurface ray-tracing method, which provides better rendering quality and performance than the built-in sampling-based approach in OSPRay. We contribute a novel reconstruction strategy-the octant method-which is continuous, adaptive and simple to implement. In this paper, we detail a comprehensive solution for interactive isosurface rendering of block-structured AMR data. However, visualizing such AMR data poses a significant challenge due to the difficulties introduced by the hierarchical representation when reconstructing continuous field values. Adaptive mesh refinement (AMR) is a key technology for large-scale simulations that allows for adaptively changing the simulation mesh resolution, resulting in significant computational and storage savings.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |